WO2001084620A1

WO2001084620A1 - Thickness measuring apparatus, thickness measuring method, and wet etching apparatus and wet etching method utilizing them

Info

Publication number: WO2001084620A1
Application number: PCT/JP2001/000351
Authority: WO
Inventors: Teruo Takahashi; Motoyuki Watanabe
Original assignee: Hamamatsu Photonics K.K.
Priority date: 2000-05-01
Filing date: 2001-01-19
Publication date: 2001-11-08
Also published as: ATE358892T1; AU2001227075A1; US6768552B2; DE60127673T2; JP2001313279A; US20030121889A1; JP4486217B2; DE60127673D1; EP1296367A1; EP1296367B1; EP1296367A4

Abstract

At each measuring time, a measuring light from a measuring light source (11) reflected from a semiconductor wafer (W) is combined with a reference light from a reference light generating section (14), and the interference light detected by a photodetector (15). A calculated thickness calculating section (16b) calculates a calculated thickness value by selecting two light intensity peaks corresponding to the upper and lower surfaces of the wafer (W) from the light intensity distribution of the intensity of interference light and the optical path length. A statistic thickness value calculating section (16c) determines a statistic thickness value by performing statistic processings including data selection, judgment if the calculated thickness is in an allowable value range, and determination of thickness variation line. A method and an apparatus for measuring the thickness of a semiconductor wafer during wet etching regardless of presence of etching liquid are realized along with a wet etching apparatus and a wet etching method utilizing them.

Description

Disclosure of the invention

Conventional thickness measuring devices for semiconductor wafers include a contact type thickness gauge and a Michelson interferometer type thickness gauge. Of these thickness gauges, the contact type thickness gauge cannot be applied to in-situ measurement. In addition, the wafer may be damaged due to contact, and high-speed measurement cannot be performed, or when a holding substrate or film is attached, the thickness cannot be measured using only the wafer. There is.

On the other hand, a Michelson interferometer-type thickness gauge measures the thickness of a semiconductor wafer without contact. As such a thickness gauge, there is an apparatus disclosed in Japanese Patent Application Laid-Open No. H5-2488817. In this apparatus, a semiconductor wafer is irradiated with measurement light to reflect reflected light from the wafer surface. The time change of the thickness is measured by the change of the reflection timing. However, in this case, since only the position of the front surface is measured, it is necessary to give initial conditions of the thickness such as the position of the back surface in order to obtain the thickness. Further, in a wet etching process using an etching liquid, the measurement light is reflected by the etching liquid on the wafer surface, so that the thickness of the semiconductor wafer cannot be measured.

The present invention has been made to solve the above problems, and has been made in consideration of the above circumstances, and provides a thickness measuring apparatus, a thickness measuring method, and a thickness measuring method capable of measuring the thickness of a semiconductor wafer during execution of wet etching. The purpose of the present invention is to provide a wet etching apparatus and a wet etching method.

In order to achieve such an object, a thickness measuring apparatus according to the present invention is a thickness measuring apparatus for measuring the thickness of a semiconductor wafer during execution of wet etching using an etching solution, wherein: At each of a plurality of measurement times separated by a time interval, a measurement light source that supplies measurement light, (2) a light branching unit that branches the measurement light from the measurement light source, and (3) a light branching unit that is branched by the light branching unit One of the measurement lights is output to the semiconductor wafer to be measured, from the etching surface side where the etching solution is supplied. A light output means for irradiating; (4) a light input means for inputting reflected light reflected by an etching liquid or a semiconductor wafer; and (5) a measuring light irradiated from the light output means is branched by a light branching means. The other of the measurement light passing through a reference optical path having a variable optical path length to generate reference light having a set reference optical path length; and (6) reflection from the light input means. (7) a light detecting means for detecting the interference light from the light coupling means, and (8) a measuring time. Using a light intensity distribution indicating a correlation between the reference light path length set by the reference light generation means and the light intensity of the interference light detected by the light detection means, and using a light intensity distribution having a light intensity larger than the set threshold value. Light intensity peaks selected from the two light intensity peaks A raw thickness value calculating means for calculating a raw thickness value of the semiconductor wafer based on a difference in optical path length of a reference optical path length between peaks, (9) a measurement time after a specified time has elapsed from the first measurement time; In each case, the thickness change line is determined by linear approximation calculation for the time change of multiple raw thickness values that are valid within the set allowable numerical value range, and the statistical thickness is calculated from the thickness change line. And (10) the statistical thickness value calculating means is the first measuring time after a specified time has elapsed from the first measuring time, and is effective up to the measuring time. For the time change of the raw thickness value, the thickness change line for selection is determined by linear approximation calculation, the selection numerical range is set from the thickness change line for selection, and the raw thickness value outside the selection numerical value range is calculated. Invalidation date, including evening selection After performing the overnight selection calculation, determine the thickness change line by linear approximation calculation for the time change of the raw thickness value that is valid after the selection, and set the allowable numerical range from the thickness change line It is characterized by the following.

Further, the thickness measuring method according to the present invention is a thickness measuring method for measuring the thickness of a semiconductor wafer during the execution of wet etching using an etching solution, wherein (1) a plurality of measuring methods at predetermined time intervals. A measuring light supply step for supplying measuring light from the measuring light source at each time; (2) an optical branching step for branching the measuring light from the measuring light source; and (3) a measuring light branched in the optical branching step. One of the half An optical output step for outputting to the conductive wafer and irradiating from the etching surface side to which the etching liquid is supplied; and (4) a measuring light irradiated in the optical output step is reflected by the etching liquid or the semiconductor wafer. (5) The reference light path length is set by passing the other of the measurement light branched in the light branching step through the reference light path having a variable optical path length. A reference light generating step for generating the reference light, and (6) an optical coupling for combining the reflected light input in the light input step with the reference light generated in the reference light generating step to generate interference light (7) a light detection step for detecting the interference light combined in the light coupling step; and (8) a reference light path length set in the reference light generation step and a light detection step for each of the measurement times. Using the light intensity distribution showing the correlation with the light intensity of the interference light detected in step 2, the light intensity distribution between two light intensity peaks selected from a plurality of light intensity peaks having a light intensity larger than the set threshold value is calculated. The raw thickness value calculation step for calculating the raw thickness value of the semiconductor wafer based on the optical path length difference of the reference optical path length, and (9) setting at each of the measurement times after the specified time has elapsed from the first measurement time Statistical calculation that determines the thickness change straight line by linear approximation calculation for the time change of multiple raw thickness values that are valid within the specified allowable numerical value range and calculates the statistical thickness value from the thickness change straight line In the (10) statistical thickness value calculation step, the first measurement time after a specified time has elapsed from the first measurement time, and (1 1) valid until the measurement time Time of raw thickness Of the thickness change line for sorting by linear approximation calculation, setting of the numerical range for sorting from the thickness changing line for sorting, and a method of invalidating the raw thickness value outside the numerical range for sorting. After performing the overnight selection calculation including evening selection, the thickness change line is determined by linear approximation calculation for the time change of the raw thickness value that is valid after the selection, and the thickness change line is determined. The feature is to set the allowable numerical range from.

In the thickness measuring apparatus and the thickness measuring method described above, the reflected light reflected by irradiating the semiconductor wafer with the measuring light and the reflected light that is branched from the measuring light and passes through a predetermined optical path are used as the light.

Four The generated interference light is detected by combining the emitted light with the reference light having the reference light path length set. Then, the thickness of the semiconductor wafer during gate etching is measured from a plurality of light intensity peaks generated in the light intensity distribution of the interference light. At this time, the measurement light applied to the semiconductor wafer is reflected on the surface of the etching solution, the upper surface (etched surface) of the semiconductor wafer, the lower surface, and the like. In the light intensity distribution, light intensity peaks corresponding to those surfaces are respectively obtained. can get. Therefore, by utilizing the two light intensity peaks corresponding to the upper and lower surfaces of the semiconductor wafer selected according to a predetermined selection criterion, the thickness of the semiconductor wafer during wet etching regardless of the presence of the etchant, or The time change can be measured. In addition, since the thickness is not determined from the reflected light from the upper surface of the wafer and the initial conditions used as the reference, the reflected light from both the upper surface and the lower surface of the wafer is used. In addition, the thickness of the semiconductor wafer can always be accurately measured.

In addition, in the thickness measurement after a predetermined time has elapsed and a sufficient raw thickness value has been obtained, a fitting calculation that approximates the time change with a straight line is performed, and the obtained thickness is calculated. The statistical thickness value is calculated from the change line. As a result, it is possible to reduce the influence of the statistical variation of the raw thickness value. In addition, in the measurement of the thickness, in addition to the statistical variation, there may be a variation due to a measurement error such that a light intensity peak from the etching liquid level is not detected. On the other hand, by limiting the raw thickness value used to determine the thickness change straight line to the raw thickness value within the predetermined allowable numerical range, the raw thickness value that caused a measurement error is excluded, and the error The effect can be reduced.

In addition, when the thickness change straight line is determined for the first time after the lapse of the specified time, the fitting calculation for selecting the thickness change line and the statistics of the raw thickness value by the fitting calculation for determining the thickness change line are performed. Processing is being performed. Here, the raw thickness value selected by statistical processing at the first measurement time after the specified time has elapsed is determined

Five The thickness change line is the initial condition for statistical processing at the second and subsequent measurement times.

Therefore, a preliminary straight-line approximation calculation for data selection is performed as described above, and the raw thickness value that caused measurement errors is invalidated using the thickness change straight line for selection and the selection numerical range, and statistical processing is performed. After performing the statistical processing including the straight-line approximation calculation again, the thickness change straight line and the allowable numerical range are set, so that the effects of error variation etc. can be efficiently reduced at each measurement time after the specified time has elapsed. It is possible to reduce the cost. Note that the data selection calculation using the thickness change line for selection is not limited to one time, but it is repeated several times to exclude extra raw thickness values such as raw thickness values that have caused measurement errors. You may go without fail.

Further, according to the jet etching apparatus and method to which such a thickness measuring device and method are applied, the thickness is obtained via the etching control means on the basis of the thickness value obtained for the semiconductor wafer during the wet etching. Accordingly, it is possible to appropriately control the end of the wet etching by stopping the supply of the etching solution, or the change of the etching rate.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 is a configuration diagram showing one embodiment of a thickness measuring device and a wet etching device provided with the thickness measuring device.

2A and 2B are views showing a method for measuring the thickness of the semiconductor wafer in the wet etching apparatus shown in FIG.

FIG. 3 is a flowchart showing an embodiment of the thickness measuring method and the etching method.

Fig. 4 is a graph schematically showing an example of the raw thickness value calculated up to the specified time.o

FIG. 5 is a graph showing the first data selection calculation.

FIG. 6 is a graph showing the second data selection calculation. FIG. 7 is a graph showing the setting of the allowable numerical value range.

FIG. 8 is a graph showing the determination of the raw thickness value within the allowable numerical value range.

FIG. 9 is a graph showing determination of a thickness change line and calculation of a statistical thickness value. BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, preferred embodiments of a thickness measuring apparatus, a thickness measuring method, an etching apparatus using the same, and an etching method using the same will be described in detail with reference to the drawings. In the description of the drawings, the same elements will be denoted by the same reference symbols, without redundant description. Also, the dimensional ratios in the drawings do not always match those described.

First, the configurations of the thickness measuring apparatus and the wet etching apparatus according to the present invention will be described together with the corresponding steps of the thickness measuring method. FIG. 1 is a configuration diagram showing an embodiment of a thickness measuring device and a wet etching device including the same. This wet etching apparatus is configured to include a thickness measuring apparatus A and a normal wet etching apparatus B except for the thickness measuring apparatus A (hereinafter, this apparatus portion is simply referred to as a wet etching apparatus B). .

The thickness measuring device A irradiates the semiconductor wafer W to be measured with measurement light, and measures the thickness of the semiconductor wafer W by using the light intensity change of the reflected light from the semiconductor wafer W and the interference light of the reference light. Is a non-contact type thickness gauge configured as described above. The measurement light used for thickness measurement is supplied from the measurement light source 11 (measurement light supply step) at each of a plurality of measurement times at predetermined time intervals, and the measurement light output from the measurement light source 11 is The light is input to an optical power blur 12 composed of a fiber power blur via an input optical fiber 11a. As the measurement light source 11, it is preferable to use a low coherence light source (for example, an SLD that generates light having a wavelength of 1.3 m). As the wavelength of the measurement light, a wavelength that sufficiently transmits the semiconductor wafer W, the etchant, and the like is selected.

The optical power bra 12 functions as an optical branching unit for branching the measuring light from the measuring light source 11, and the measuring light input to the optical power bra 12 is used as a measuring optical fiber toward the measuring optical path. No. 13a and a reference optical fiber 14a directed to the reference optical path are branched (light branching step). The branched measurement lights are input to a probe head 13 for measuring the thickness and a reference light generation unit 14 for generating the reference light, respectively.

The probe head 13 is a light output means for irradiating the measurement light onto the semiconductor wafer W, and a light output means for re-inputting the reflection light which is reflected by the semiconductor wafer W or the etching liquid. This is an optical input / output unit that functions as an optical input unit. Of the light branched by the optical power blur 12, the measurement light branched to the optical fiber 13 a side is output from the probe head 13 to the semiconductor wafer W and is directed upward to the semiconductor wafer W. Light is emitted from the etching surface on the surface side (light output step). As the measurement light, light having a wavelength sufficiently transmitting through the semiconductor wafer W or the like is used as described above, but part of the light is reflected at each interface, and the reflected light is returned to the probe again. The data arrives at input 13 and is input (optical input step). The reflected light that has reached the probe 13 and has been re-input is input to the optical power bra 12 via the optical fiber 13 a.

On the other hand, in the reference light generation section 14, reference light for measuring the thickness (optical path length) is generated by interference light with the reflected light from the semiconductor wafer W or the like (reference light generation step). The measurement light branched to the optical fin 14a by the optical power bra 12 is transmitted to the optical path length modulation optical system disposed between the output end of the optical fiber 14a and the reflection mirror 14c. The reference light passes through the reference optical path 14b and becomes the reference light in which the optical path length (reference optical path length) of the reference light with respect to the optical path length (reflected optical path length) of the reflected light from the semiconductor wafer W or the like is set.

In the present embodiment, the measurement light output from the output end of the optical fiber 14a transmits through the parallel flat glass substrate 14d, reaches the reflection mirror 14c, and is reflected. The reflected light from the reflecting mirror 14c again passes through the glass substrate 14d in the reverse direction, and is used as a reference light having an appropriate reference optical path length set through the optical fin 14a. Entered in 1 2

The above-described reference light generator 14 is configured such that the optical path length of the reference optical path 14 b is variable. You. That is, the glass 14f on the reference optical path 14b is attached to the galvanometer 14e. The galvanometer 14 e operates based on a periodic signal from the reference optical path length control unit 17, whereby the inclination of the glass substrate 14 d with respect to the reference optical path 14 b periodically changes. Change. At this time, since the thickness of the glass substrate 14 d as viewed in the direction of the reference optical path 14 b changes, the optical path length of the reference optical path 14 b periodically changes, and the reflection occurs. The reference optical path length with respect to the optical path length (the timing of the reference light with respect to the reflected light) is periodically scanned.

The optical power bra 12 is a light branching means for branching the measurement light from the measurement light source 11 as described above, and also reflects the reflected light from the probe head 13 and the light from the reference light generation unit 14. It also functions as an optical coupling means for coupling the reference light. The reflected light reflected by the semiconductor wafer W and the like and returned to the probe head 13 and input thereto, and the reference light having the reference light path length set in the reference light generation unit 14 are combined by the optical power blur 12. The light becomes interference light (optical coupling step), and is input to a photodetector 15 such as a photodiode (PD) via an output optical fiber 15a and detected (light detection step).

At each measurement time, the data of the interference light detected by the photodetector 15 is processed by the thickness calculator 16 and the thickness of the semiconductor wafer W is calculated based on the data. (Thickness calculation step).

The detection signal obtained by detecting the interference light by the photodetector 15 is input to the raw thickness value calculation unit 16b via the signal processing circuit 16a of the thickness calculation unit 16. The detection signal from the photodetector 15 provides a light intensity of the interference light. Similarly, the angle signal of the galvanometer 14 e (glass substrate 14 d) from the reference light path length control unit 17 is also transmitted to the raw thickness value calculation unit 16 b via the signal processing circuit 16 a. Has been entered. From this angle signal, the reference optical path length in the reference optical path 14b or the amount of change in the optical path length can be obtained.

In the raw thickness value calculation unit 16b, at each measurement time, the change in the light intensity of the interference light due to the reference light path length (correlation) from the light intensity data and the reference light path length data. Is created. Then, in the obtained light intensity distribution, a plurality of light intensity peaks are specified as those having the peak intensity exceeding the set threshold intensity, and two light intensity peaks selected from the light intensity peaks are identified. The raw thickness value of the semiconductor wafer W is calculated using the peak (raw thickness value calculation step).

The raw thickness value calculated by the raw thickness value calculation unit 16 b is further calculated by the statistical thickness value calculation unit 1

6 Entered in c. In the statistical thickness value calculation unit 16c, a thickness change straight line is determined by a linear approximation calculation with respect to the time change of a plurality of raw thickness values, and a statistical thickness value is calculated (statistical thickness value calculation step).

The details of the method of calculating the reflected light from the semiconductor wafer W and the like, the light intensity peak of the light intensity distribution generated corresponding thereto, the raw thickness value, and the statistical thickness value will be described later.

The wet etching apparatus B performs wet etching of one surface (the upper surface in FIG. 1; hereinafter, referred to as an etching surface) of a semiconductor wafer W to be subjected to the etching process (the measurement object of the thickness measuring device A) by an etching solution. It is composed of

The semiconductor wafer W is fixed on a turntable 22 while being held by a holding substrate 21 made of a glass substrate or the like disposed on the side opposite to the etching surface. The turntable 22 is driven to rotate by a rotation drive unit 23, whereby the semiconductor wafer W is rotated during the gate wake-up. When the semiconductor wafer W is provided with a pattern, the surface with the pattern is the holding substrate 21 side, and the wet etching is performed with the surface opposite to the pattern as the etching surface.

The supply of the etching liquid to the etching surface of the semiconductor wafer W is performed by the etching liquid supply unit 24. The etching liquid supply unit 24 supplies and stops an etching liquid to the semiconductor wafer W, or supplies cleaning water. When the etching liquid is supplied from the nozzle 24 a to the etching surface of the rotating semiconductor wafer W by the etching liquid supply unit 24, the supplied etching liquid is supplied to the semiconductor wafer W.

Ten A thin etching liquid layer E is formed on the surface of W, and the surface of the semiconductor wafer W is etched by the etching liquid layer E.

The rotation table 22, the holding substrate 21 placed on the rotation table 22, and the rotation of the semiconductor wafer W by the rotation drive section 23, and the etching liquid supply section 24 to etch the semiconductor wafer W on the etching surface Supply and stop of the solution or the cleaning solution are controlled by the etching control unit 25.

The probe head 13 of the thickness measuring device A is positioned at a position facing a predetermined portion of the etching surface of the semiconductor wafer W mounted on the turntable 22 together with the holding substrate 21, and at a position corresponding to the etching surface. It is installed so that the optical path of the measurement light irradiated toward it is almost perpendicular to the etching surface. At this time, the reflected light, in which the vertically irradiated measurement light is reflected by the semiconductor wafer W or the like, is efficiently input to the probe head 13 again.

In addition, in order to prevent corrosion of the lens and the like due to the scattered etching solution, it is preferable that the probe head 13 be provided with a transparent sheet such as a salted vinyl, which is resistant to an etching solution, as a protective film. . Alternatively, a cylinder may be attached to the tip of the probe head 13 and the inside thereof may be pressurized to prevent the etchant from adhering. Here, a description will be given of an example of a wet etching method of a semiconductor wafer W using the etching apparatus of FIG. 1 including the thickness measuring apparatus A and the etching apparatus B.

First, the semiconductor wafer W held on the holding substrate 21 is placed on the turntable 22. Then, the rotation drive of the turntable 22 is started based on the instruction signal from the etching control unit 25. Subsequently, the supply of the etchant to the etching surface of the semiconductor wafer W is instructed to the etchant supply unit 24, and the etching of the semiconductor wafer W is started (etching start step).

When the etching starts, the thickness of the semiconductor wafer W is measured by the thickness measuring device A and the thickness measuring method described above (thickness measuring step). Thickness measurement is performed at the measurement time specified by the operator or at a preset time interval.

11 It is performed automatically at the measurement time. Then, from the thickness data obtained at each measurement time, the raw thickness value is calculated in the raw thickness value calculation unit 16b. In addition, in the thickness measurement after a sufficient number of thickness measurements have been performed after the specified time has elapsed, the statistical thickness value calculation unit 16c calculates the time change of the thickness from the raw thickness value calculated at each measurement time. And a statistical thickness value subjected to statistical processing is calculated.

Then, the thickness of the semiconductor wafer W and its time change in the wet etching process being performed are evaluated from the thickness change straight line and the statistical thickness value. Regarding the evaluation of the thickness, for example, the thickness can be automatically evaluated in the thickness calculating unit 16 of the thickness measuring device A. Alternatively, a configuration may be adopted in which a display device (display) is connected to the thickness calculator 16 and the display device displays the thickness data, and the operator evaluates based on the displayed data. .

At the end time of the etching, the supply of the etching liquid by the etching liquid supply unit 24 is stopped by the instruction signal from the etching control unit 25. Subsequently, cleaning water is supplied to the etched surface of the semiconductor wafer W for a predetermined time, and the semiconductor wafer W is cleaned. After the supply of the cleaning water is stopped and the cleaning of the semiconductor wafer W is completed, the turntable 22 is driven to rotate for a predetermined time to remove the cleaning water from the etched surface of the semiconductor wafer W. Then, when the removal of the cleaning water is completed, the rotation of the turntable 22 by the rotation drive unit 23 is stopped, and the entire etching process of the semiconductor wafer W is completed (etching end step).

At this time, the end time of the wet etching may be determined based on the etching time or the etching rate data given in advance, but the raw thickness value measured by the thickness measuring device A may be determined. It is preferable to calculate and use the end time at which the set end point thickness is obtained from the thickness change line obtained overnight (end time calculation step).

The calculation of the end time may be made automatically by the statistical thickness value calculator 16c, or may be determined by the operator based on the data displayed on the display device.

12 Noh. When the end time is obtained by the statistical thickness value calculating section 16c, an end instruction signal indicating the end time is output from the statistical thickness value calculating section 16c of the thickness calculating section 16 and the end instruction is given. The etching control unit 25 can be configured to perform the end control of the gate etching based on the signal.

A method of measuring and calculating a raw thickness value of the semiconductor wafer W at each measurement time by the thickness measuring device A and the thickness measuring method of the above embodiment will be described. 2A and 2B are diagrams schematically showing a method for measuring the thickness of the semiconductor wafer W in the gate etching apparatus shown in FIG. 1, and FIG. 2A is a diagram showing the measurement on the semiconductor wafer W. FIG. 2B is a side cross-sectional view showing light irradiation and re-input of reflected light to the probe head 13, and FIG. 2B is a graph showing a light intensity distribution of interference light obtained in the photodetector 15. You. In FIG. 2A, the optical path of the measurement light applied to the semiconductor wafer W and the optical path of the reflected light to the probe head 13 are shown with their positions shifted for the sake of clarity. It is.

The measurement light L◦ branched from the optical power bra 12 and output from the probe head 13 passes through the etchant layer E, the semiconductor wafer W, and the holding substrate 21 sequentially, and A part of the measurement light L0 is reflected at each interface between the adjacent layers. That is, the reflected light L1 from the surface of the etchant layer E, the reflected light L2 from the upper surface of the semiconductor wafer W, the reflected light L3 from the lower surface of the semiconductor wafer W, and the reflected light L3 from the lower surface of the holding substrate 21. The light L4 is reflected respectively, returned to the probe 13 and returned to the probe.

The re-input reflected lights L 1 to L 4 are passing through different reflected light path lengths depending on the reflected surface as shown in FIG. The timing of input to the photodetector 15 via the bra 12 differs. On the other hand, the reference light generation unit 14 periodically changes the optical path length of the reference light path 14b as described above, and the reference light path length (the evening of the reference light with respect to the reflected light) is scanned. On.

13 At this time, if the optical path length from the optical power blur 12 to each interface where the reflected light L1 to L4 is reflected and the optical path length from the optical power blur 12 to the reflection mirror 14c match, the optical path The reflected light and the reference light having the same length and timing are strengthened by interference, and the photodetector 15 detects interference light having a large light intensity.

FIG. 2B shows a light intensity distribution showing a correlation between the reference light path length (optical path length change amount) obtained by scanning the light path length and the interference light intensity in correspondence with the cross-sectional view of FIG. 2A. In this graph, one axis indicates the optical path length change amount of the scanned reference optical path 14b, and the other axis indicates the light intensity of the interference light detected by the photodetector 15. Note that the reference optical path length (optical path length change amount) and the optical path length difference do not necessarily correspond to the respective thicknesses due to the difference in the refractive index of each of the etchant layer E, the semiconductor wafer W, and the holding substrate 21. In FIG. 2A and FIG. 2B, for the sake of explanation, it is assumed that there is no difference in the refractive index, and the sectional view and the graph are shown in correspondence.

As shown in this graph, scanning in the direction of increasing the optical path length change amount from the smaller one (increase the reference optical path length) corresponds to the reflected light L1 from the surface of the etchant layer E. Light intensity peak P 1 (liquid level peak P 1), light intensity peak P 2 corresponding to the reflected light L 2 from the upper surface (etched surface) of semiconductor wafer W (wafer upper surface peak P 2), and semiconductor wafer W The light intensity peak P 3 corresponding to the light L 3 reflected from the lower surface (peak P 3 on the lower surface of the wafer) and the light intensity peak P 4 corresponding to the light L 4 reflected from the lower surface of the holding substrate 21 (peak P 4 on the lower surface of the substrate). ) Are sequentially obtained.

These light intensity peaks: For P1 to P4, set an appropriate light intensity threshold (threshold) for the light intensity distribution, and exclude extra peaks such as small light intensity peaks due to noise signals. Specified. In FIG. 2B, the light intensity Pt is indicated by a dotted line as such a threshold intensity.

Also, the optical path length range over which the light intensity peak is scanned can be set by the scan range of the optical path length in the reference optical path 14 b in the reference light generation unit 14.

14 If necessary, an optical path length range used for specifying a light intensity peak may be selected and set from the scanned optical path length range. Such selection of the optical path length range may be given to the thickness calculating section 16 in advance, or the operator may select the optical path length range from the light intensity distribution displayed on the display device connected to the thickness calculating section 16. It is also possible to select and give an instruction by operating the mouse cursor.

In the thickness measurement, a plurality of light intensity peaks are specified by applying the above-described condition of the threshold light intensity or the condition of the optical path length range to the obtained light intensity distribution. Then, from these light intensity peaks, two light intensity peaks corresponding to the reflected light from the upper surface and the lower surface of the wafer are selected based on a predetermined selection criterion.

Here, as for the above light intensity peaks P 1 to P 4, their light intensity ratios and the like vary depending on the state of the semiconductor wafer W etching liquid layer E and the like, but the order with respect to the optical path length variation does not change. . For example, the state of the etchant layer E changes depending on how the etchant flowing out of the nozzle 24a flows on the etching surface. At this time, the angle of the etchant layer E surface with respect to the optical path of the measurement light is changed. Therefore, the light intensity of the reflected light L1 reaching the probe 13 from the surface of the etching solution layer E to the probe also changes. The light intensity ratio also varies depending on the material used as the semiconductor wafer W (such as Si, GaAs, and DopSi), and the material of the holding substrate 21.

On the other hand, regarding the optical path length, even when the state such as the light intensity changes as described above, the order of the light intensity peaks P1 to P4 with respect to the optical path length does not change. Accordingly, by selecting two light intensity peaks for the obtained light intensity peaks using the order of the light intensity peaks as a selection criterion, the light intensity peaks P corresponding to the upper and lower surfaces of the wafer are obtained. 2, P3 can be selected.

Then, in the light intensity distribution shown in FIG. 2B, the optical path length difference between the second light intensity peak P 2 and the third light intensity peak P 3 from the smaller reference optical path length is the same as that of the semiconductor wafer W. This corresponds to the optical path length difference from the upper surface to the lower surface. Therefore, these two

Fifteen The raw thickness value of the thickness of the semiconductor wafer W can be calculated from the optical path length difference between the light intensity peaks P2 and P3 of FIG.

In particular, rather than measuring the optical path length and its time change with respect to one light intensity peak, the measurement method using the two light intensity peaks P 2 and P 3 as described above reduces the thickness of the semiconductor wafer W. It can measure more directly and correctly. Further, the thickness measurement during the execution of the wet etching in which the etching liquid is flowing on the etching surface of the semiconductor wafer W becomes possible regardless of the presence of the etching liquid.

The optical path length difference between the light intensity peaks P 2 and P 3 described above corresponds to the optical thickness of the semiconductor wafer W. Therefore, the final green thickness value is obtained by dividing the obtained optical path length difference by the refractive index of the semiconductor wafer. As the value of the refractive index of the semiconductor wafer w used for calculating the raw thickness value, the value may be used if the refractive index is known. If necessary, it is preferable to measure the refractive index in advance on a wafer whose thickness has been measured by another method using a micro gauge, a microscope, or the like, and use that value.

As the predetermined selection criterion and I ^ for selecting the two light intensity peaks P2 and P3 corresponding to the upper and lower surfaces of the wafer, specifically, some selection methods can be applied. For example, there is a method of setting the optical path length range used for calculating the raw thickness value to the range R1 or R2 (FIG. 2B), and selecting the second and third light intensity peaks from the top. Alternatively, there is a method of setting the optical path length range to the range: R 1 (range R 2) and selecting the first and second from the bottom (second and third from the bottom).

Regarding the light intensity peak P4 corresponding to the lower surface of the substrate, the reflected light L4 may not be detected with a sufficient intensity due to a long optical path length due to the thickness of the holding substrate 21 and the like. In such a case, in order to stably identify the light intensity peak, the light path length range R1 excluding the light path length range where the light intensity peak P4 is detected is changed to the light intensity peak selection. It is preferable to set it. Further, even when the holding substrate 21 is not used, the optical path length range R 1 is set in the same manner.

16 As described above, a non-contact type thickness measuring device capable of measuring the thickness of the semiconductor wafer W during the execution of the gate etching, and a liquid jet etching device and a liquid jet etching method including the same are obtained.

Note that the state of the etching solution layer E changes as described above, and the thickness thereof also changes with time similarly to the surface angle. This changes the optical path length difference between the light intensity peaks Pl and P2. At this time, not only does the peak position of the light intensity peak P1 shift, but also the refractive index of the etching liquid layer E. As a result, the optical path length from the probe head 13 to the semiconductor wafer W changes. Therefore, the peak positions of the light intensity peaks P 2 and P 3 are similarly shifted. Also in this case, the light intensity distribution corresponding to the portion below the upper surface of the semiconductor wafer W (the semiconductor wafer W and the holding substrate 21) shifts by the same amount as a whole, and thus the optical path lengths of the light intensity peaks P2 and P3. Each optical path length difference such as the difference is not affected by the shift of the peak position.

When a pattern is formed on the surface of the semiconductor wafer W opposite to the etched surface, if the beam diameter of the measurement light is smaller than the pattern, the thickness at each part of the pattern is larger than the pattern. Also, if the beam diameter is large, an average thickness within the beam range can be obtained. Further, in the wet etching apparatus shown in FIG. 1, since the semiconductor wafer W is rotated during the etching, the average thickness is measured in the thickness measurement in this case.

Statistical processing of the raw thickness value obtained by the above-described thickness measurement method and raw thickness value calculation method will be described. The thickness value (raw thickness value) of the semiconductor wafer W calculated by the thickness measurement at each of a plurality of measurement times at predetermined time intervals is as follows: (1) statistical variation (statistical variation); 2) There is a value variation due to two causes: variation due to measurement error (error variation). Among them, (1) Statistical variation is inevitable even in the thickness measurement performed correctly, and the variation in the raw thickness value is within the allowable range as a whole.

On the other hand, (2) error variation occurs due to, for example, the following causes. sand

17 That is, the angle of the surface of the etching liquid layer E with respect to the measurement light changes depending on the flow of the etching liquid as described above, and accordingly, the light intensity of the reflected light L1 from the surface of the etching liquid layer E changes. . In particular, when the surface of the liquid rises due to the undulation of the surface, the angle of the reflected light L1 with respect to the measurement light increases, so that the reflected light L1 is not input to the probe head 13.

At this time, in the light intensity distribution of FIG. 2B, if the light intensity of the liquid level peak P1 is smaller than the threshold value Pt, it is not specified as the light intensity peak. Also, depending on the measurement conditions, the wafer upper surface peak P2 and the wafer lower surface peak P3 may not be similarly specified as the light intensity peak.

In such a case, since each light intensity peak is incorrectly assigned to each surface, the raw thickness value is incorrectly calculated. For example, when selecting the second and third light intensity peaks from the top in the optical path length range R2, a measurement error occurs in which the thickness of the holding substrate 21 is calculated as the raw thickness value. When selecting the first and second light intensity peaks from the bottom in the optical path length range R1, the thickness of the etching solution layer E or the total thickness of the etching solution layer E and the semiconductor wafer W is determined by the raw thickness. A measurement error calculated as a value occurs. Also, when a peak due to a noise signal exceeding the threshold level Pt occurs between the light intensity peaks P1 to P4, the noise peak is regarded as the above light intensity peak, and an incorrect thickness is obtained. Can be

With respect to the variation of the raw thickness value due to these statistical variations and error variations, the thickness measurement method and the etching method in the etching apparatus shown in FIG. Eliminating or reducing the effect of each value variation by selecting the data by applying the tolerance range in the evening and determining the thickness change line by performing a linear approximation calculation on the selected raw thickness value data I have. In other words, the effect of error variation is removed by selecting raw thickness value data, and the effect of statistical variation is reduced by linear approximation calculation.

18 Hereinafter, a method of determining a thickness variation straight line from the raw thickness values calculated at each measurement time, a method of calculating a statistical thickness value, and a method of determining the end time of wet etching will be specifically described. FIG. 3 is a flowchart showing one embodiment of a thickness measuring method and a wet etching method in the wet etching apparatus shown in FIG.

In the present etching apparatus, when the etching of the semiconductor wafer W is started by the etching apparatus B, the thickness is measured at each of a plurality of measurement times t at predetermined time intervals. The measurement is performed (Step S101). The time interval for instructing the measurement time is set appropriately according to the etching time and the etching rate, for example, a 5 Hz time interval for the entire etching time of 1 to 2 minutes, and automatically set at each measurement time. It is preferable to perform the thickness measurement at first. In addition, a fixed time interval may be used for the entire etching time, or a different time interval may be used.

When the measurement light is supplied from the measurement light source 11 and the thickness measurement is performed, each data from the photodetector 15 and the reference optical path length control unit 17 is generated via the signal processing circuit 16 a of the thickness calculation unit 16. The value is input to the thickness value calculator 16b.

Next, the light thickness distribution (see FIGS. 2A and 2B) is created in the raw thickness value calculation unit 16b, and the raw light intensity distribution at the measurement time t is calculated using the two selected light intensity peaks. The thickness value RTh (t) is calculated (S102). Here, if the number of light intensity peaks specified in the light intensity distribution is less than two, the raw thickness value cannot be calculated, so that RTh (t) = 0 m and the raw thickness value is invalidated.

Furthermore, it is determined whether or not the classification based on the number of the specified light intensity peaks is ON for the calculated raw thickness value (S103). Classification is performed (S104). In the classification based on the number of peaks, the light identified as a peak having a light intensity larger than a threshold value within the optical path length range set for the light intensity distribution, out of the calculated raw thickness value RTh (t). Number of intensity peaks

19 If there are less than three, the raw thickness value is invalidated.

If the number of identified peaks is less than 3, one of the light intensity peaks P1, P2, and P3 on the liquid surface, wafer upper surface, and wafer lower surface is detected. This corresponds to the case where an unmeasured error occurs, and it is highly likely that an incorrect raw thickness value is obtained. Therefore, by performing the above-described classification based on the number of peaks, it is possible to exclude at least a part of the raw thickness value accompanying the measurement error and reduce the influence of the error variation. Such classification based on the number of peaks is particularly effective when the thickness of the liquid layer E and the thickness of the wafer W are almost the same, and it is difficult to judge the occurrence of measurement errors by applying the selection numerical range and the allowable numerical range described later. effective.

However, when a sufficient effect can be obtained by statistical processing such as data selection, which will be described later, such as when the thickness of the wafer W is sufficiently thicker than the liquid layer E, such classification based on the number of peaks is performed. You don't have to. In this case, all raw thickness values other than data for which the number of light intensity peaks is less than two and for which raw thickness values cannot be calculated are valid.

Next, it is determined whether or not the time elapsed from the first measurement time is equal to or longer than a specified time (S105), and if less than the specified time, the execution of the thickness measurement and the calculation of the raw thickness value are repeated. On the other hand, if the specified time has been reached, statistical processing on the raw thickness value is started. This specified time is a time for judging whether or not a statistical score sufficient for the raw thickness value of the semiconductor wafer W to be evaluated has been obtained, and the elapsed time (time width) from the first measurement time. It is preferable to specify the Alternatively, it is also possible to specify the specified time based on the number of statistical points at which the thickness was measured. In the following, the specified time is specified by the time width regardless of the statistical score, and the time width Tc from the first measurement time is defined as the specified time.

If it is determined that the time from the start of the thickness measurement has reached the specified time Tc, then it is determined whether the allowable numerical range for the raw thickness value has been set (S106). If the allowable numerical value range is not set, the allowable numerical value range is set since the time is the first measurement time after the specified time Tc has elapsed from the first measurement time.

20 The method of setting the allowable numerical value range at the first measurement time (S107 to S109) will be described with reference to graphs schematically shown in FIGS. Here, in the graphs of FIGS. 4 to 9 shown below, the horizontal axis represents the etching time t (= measurement time t), and the vertical axis represents the thickness Th of the semiconductor wafer W at each time. I have. Also, the raw thickness value RTh calculated at each measurement time t shown in each graph

As for (t), those that are valid at each stage of the graph are indicated by black circles. On the other hand, those that are invalid are shown in white circles, or are omitted from the graph for easy viewing. In addition, the end thickness ThO of the semiconductor wafer W set in advance as a target of wet etching is indicated by a dotted line parallel to the horizontal axis.

Fig. 4 shows the distribution and time change of the raw thickness value measured and calculated up to the measurement time tn at the first measurement time tn when tn = Tc and the specified time Tc was reached. 5 is a graph showing an example of the above. Of the raw thickness values at each measurement time, among the raw thickness values at R Th (t) = 0〃m, the raw thickness values at three days are two identified light intensity peaks. It is a night (white circle) that is invalid for less than one day. The setting of the permissible numerical value range is performed using other valid data (black circles).

As shown in Fig. 5, the first de-night selection calculation is performed for the time change of the effective raw thickness value RTh (t) indicated by the black circle in Fig. 4 (S107). First, a linear approximation calculation (for example, fitting calculation such as a least squares method) is performed on the data of the thickness values RTh (t), and the first thickness change line FThl ( t) is determined. This thickness change straight line FTh 1 (t) is a straight line for overnight selection. Subsequently, the first selection numerical range DTh1 from the thickness change line FTh1 (t) for selection is set. In the present embodiment, first, a variation value 1 of the raw thickness value RTh (t) is calculated with respect to the thickness change straight line F T1 (t). On the other hand, a screening constant D The 1 for obtaining the screening numerical range DTh 1 is set in advance. The numerical range selected from these values, D Thl, is DThl = lxDThc l

twenty one Is set. As the variation 1 of the raw thickness value, for example, a standard deviation value from a thickness change straight line for selection in the raw thickness value can be used.

According to the set sorting numerical range D Th1, the first de-sorting for the raw thickness value is performed. That is, the raw thickness value within the range from the thickness change straight line FThl (t) to the soil DThl (in FIG. 5, two broken lines sandwiching the thickness change straight line FTh 1 (t) for sorting from above and below) In addition to the above, the data shall remain valid, and the data of the raw thickness value outside this range (the 5 day data points indicated by white circles in Fig. 5) shall be deasserted.

Next, as shown in FIG. 6, a second data selection calculation is performed for the time change of the effective raw thickness value R Th (t) indicated by a black circle in FIG. 5 (S108). The method of data selection calculation is almost the same as the first one. That is, a straight-line approximation calculation is performed on the raw thickness value RTh (t) to determine a second thickness change straight line FTh2 (t) for the second sorting, which is an approximate line.

Subsequently, the second selection numerical range DTh2 from the thickness change straight line F Tli 2 (t) for selection is similarly set as DTh2 and 2 xDThc 2 from the variation value 2 and the selection constant DThc 2. . Then, the value of the raw thickness value within the range of soil DTh2 from the thickness change straight line FTh2 (t) remains valid, and the value of the raw thickness value outside this range (shown by a white circle in Fig. 6). Is invalidated). This is the end of the overnight selection calculation of the raw thickness value in the present embodiment.

Next, the raw thickness value that has been selected and validated by the data

From the time change of the data of RTh (t), the allowable numerical value range DTh is set as shown in FIG. 7 (S109). First, a straight-line approximation was made to the raw thickness values that were validated in the first and second de-night selection calculations (the 8 de-night points indicated by black circles in Figures 6 and 7). The calculation is executed to determine a thickness change straight line FTh _tn (t) which is an approximate straight line indicating a time change of the raw thickness value. Here, the subscript tn indicates that the thickness change straight line is the thickness change straight line determined at the measurement time tn.

twenty two Subsequently, an allowable numerical range DTh from the thickness change straight line FTh _tn (t) is set. The method of setting the allowable numerical value range DTh in the present embodiment is almost the same as the method of setting the above-described selection numerical value ranges DThl and DTh2. That is, first, the thickness changes linearly FTli. Relative _tn (t), calculates the variation value beauty, such as the standard deviation of effective and has been and raw thickness value RTh (t) at this stage. On the other hand, an allowable constant DT hc for obtaining the allowable numerical range DTh is set in advance. From these values, the allowable value range DTh is set as DTh = xDThc.

The allowable numerical range DTh set here is used to determine whether the raw thickness value is valid or invalid at each subsequent measurement time (the second or later measurement time after the specified time Tc has elapsed from the first measurement time). Used. After the determination of the thickness change line F Th _tn (t) at the measurement time tn and the setting of the allowable numerical value range DTh are completed, the process proceeds to execution of the next thickness measurement and calculation of the raw thickness value.

At the second and subsequent measurement times after the specified time Tc has elapsed from the first measurement time, the allowable numerical value range has already been set, so the above-mentioned data selection and the (re) setting of the allowable numerical value range are not performed. Perform judgments inside and outside the allowable range. The determination method (S110 to S112) of the inside and outside of the allowable range at the second and subsequent measurement times will be described with reference to the graphs schematically shown in FIGS.

With respect to the raw thickness value RTh (tm) calculated at the second and subsequent measurement times tm, as shown in FIG. 8, a judgment is made as to whether the raw thickness value is outside the allowable range (S110). Whether it is within the allowable range or outside the range is determined from the thickness change straight line FT h _tn (t) determined at the previous measurement time (here, measurement time tn). It is determined by whether it is within.

That is, the thickness change straight line FTh _tn (t) determined at the previous measurement time tn is excluded (dotted line), and the expected thickness FT h _tn (tm) at the current measurement time tm is obtained. Then, the raw thickness value RTh (tm) obtained by the thickness measurement performed at the measurement time tm is in the range from FTh _tn (tm) to the soil DTh (in FIG. 8, the thickness change straight line FTh _t

twenty three _n (t) is shown by the two broken lines sandwiching it from above and below), the value of the raw thickness value RTh (tm) is validated. On the other hand, if it is out of the range, the value of the raw thickness value is invalidated. Then, based on the determination result, the thickness change straight line _FThtm (t) for the current measurement time tm is determined (S111).

In the graph of Fig. 8, the raw thickness value RTh (tm) is

This shows a case where the value is within the allowable numerical range DTh from Th _tn (t). At this time, the raw thickness value RTh (tm) is valid. Then, as shown in Fig. 9, a straight-line approximation is performed on the data of the effective raw thickness value (black circle) that includes this raw thickness value RTh (tm) and is within the time range of the specified time Tc from the measurement time tm. Perform the calculation to determine a new thickness change line, _FThtm (t).

On the other hand, if the raw thickness value RTh (tm) is out of the allowable numerical range DTh from the previous thickness change line FT h _tn (t), the raw thickness value; Th (tm) is invalidated. You. In this case, without performing the linear approximation calculation, as it is the last thickness change linearly, that determine the current measurement time thickness variation in tm linear _{FT h tm (t) = FTh} tn (t).

When the thickness change linearly FTH _tm (t) is determined, the statistical thickness value S Th at the measurement time tm the (tm), is calculated by _{STh (tm) = FTh tm (} tm) (S 11 2), the measurement The statistical processing of the day at time tm is ended. Then, it is determined whether or not the calculated statistical thickness value STh (tm) has reached the end-point thickness ThO (S113). If the statistical thickness value STh (tm) is equal to or less than the preset gate etching end thickness ThO, the end signal is output from the thickness calculating unit 16 (statistical thickness calculating unit 16c) to the etching control unit 25. To complete the gate etching. On the other hand, as shown in FIG. 9, if the end point thickness ThO has not been reached, the wet etching process is continued and the next thickness measurement is performed.

Thickness measuring device, thickness measuring method, and wet etching using the same

twenty four In the apparatus and the wet etching method, the reflected light generated from the measurement light and the interference light of the reference light are detected, and the light intensity peaks corresponding to the upper surface and the lower surface of the semiconductor wafer are specified by the light intensity distribution with respect to the change in the optical path length. Select and calculate the raw thickness value from the optical path length difference. Thus, the thickness of the semiconductor wafer can be measured during the gate etching regardless of the presence of the etching solution. Also, since two light intensity peaks from both the upper and lower surfaces of the wafer are used, the thickness of the semiconductor wafer can be accurately measured even when the state of the etching liquid layer changes.

In addition, at the measurement time when there is a sufficient number of statistics of the raw thickness value after the specified time has elapsed, statistical processing including determination of outside the allowable numerical value range and determination of the thickness change line by straight-line approximation calculation is performed. I have. Thereby, the statistical thickness value in which the influence of the error variation and the statistical variation is sufficiently reduced can be calculated as the thickness of the semiconductor wafer W at the measurement time.

In particular, at the first measurement time after the lapse of the specified time, before determining the thickness change straight line, a preliminary linear approximation calculation for data selection, setting of the selection numerical range, and data including data selection are performed. After the overnight selection calculation is performed, the thickness change line is determined and the allowable numerical value range is set based on the raw thickness value that is valid after the end of the overnight selection calculation. As a result, it is possible to obtain the thickness of the semiconductor wafer W more accurately by efficiently and efficiently excluding the influence of error variation and the like from the thickness variation line and the statistical thickness value. At the second and subsequent measurement times, the raw thickness value is determined from the set allowable numerical value range and the previous thickness change line, and a new straight-line approximation calculation is performed only when the raw thickness value is valid. By doing so, efficient calculation of statistical thickness values has been realized.

Regarding the setting of the permissible numerical value range and the selection numerical value range at the first measurement time after the lapse of the specified time, in both cases, the variation value from the thickness change straight line of the raw thickness value effective at that stage was determined in advance. Set based on the allowable constant and the selection constant

twenty five I have. As a result, the setting of the permissible numerical value range and the selection numerical value range reflects the state of the data of the actual raw thickness value that is the target, so it is possible to set a numerical value range suitable for the data it can.

Here, the dispersion of the raw thickness value decreases every time the selection is performed, so that the allowable numerical value range is usually set to a smaller value range than the selection numerical value range. That is, in the above-described example, the first selection numerical range DTh1, the second selection numerical range DTh2, and the allowable numerical range DTh usually satisfy DThl> DTh2> DTh. However, since the magnitude relation of such a numerical range is mainly determined by the variation value, such a magnitude relation is not always obtained by the allowable constant and the sorting constant excluding the variation value.

Regarding the permissible numerical value range and the selected numerical value range, if the variation value is predicted in advance, the operator, etc., directly specifies the numerical value range without specifying the allowable constant and the selection constant. May be set. Even in this case, it is preferable that the respective numerical ranges are set so as to satisfy the magnitude relation described above. Also, if necessary, set some numerical ranges (for example, the second screening numerical range and the final allowable numerical range) equal, or set the allowable numerical range slightly wider than the screening numerical range. Is also good. ,

In the straight-line approximation calculation at the second and subsequent measurement times after the specified time has elapsed, the raw thickness value before the specified time range from the measurement time is invalidated and is not used for determining the straight line. Such a range setting is effective in responding to the time change when the etching rate changes over time during wet etching, etc., and ensures a more accurate score while securing a sufficient number of statistics. The thickness change line and the statistical thickness value can be obtained. However, if the change in the etching rate is not a problem, the raw thickness value from the first measurement time may always be used. In this case, as the time elapses, the statistical score of the raw thickness value used for the statistical processing increases.

26 In the jet etching apparatus and the jet etching method of FIG. 1 to which the above-described thickness measuring apparatus is applied, the etching control section 25 is based on the thickness variation straight line and the statistical thickness value obtained by the statistical thickness calculating section 16c. Through this, it is possible to appropriately control the end of the etching by stopping the supply of the etching liquid by the etching liquid supply unit 24 or the change of the etching rate. In particular, regarding the thickness of the semiconductor wafer obtained after the end of wet etching, the end time is determined based on the thickness change line and the statistical thickness value and the preset end thickness, so that the end thickness can be calculated from the end thickness. This can suppress variations in the semiconductor device, and can improve the efficiency of semiconductor device manufacturing and improve the yield.

More specifically, in the above example, if the calculated statistical thickness value is equal to or less than the end point thickness, the measurement time is set to the end time, and the thickness calculation unit 16 (statistical thickness value calculation unit 16 c ) Outputs an end instruction signal to the etching control section 25 to end the etching.

In addition to the above, a configuration is also possible in which the end time is predicted using the thickness change straight line. That is, as shown in FIG. 9, the thickness change line FT h _tm (t) is extended (extended) to find an intersection with the straight line indicating the end point thickness, and the time te at the intersection is defined as the end time. Can be predicted. When the end time is predicted in advance in this way, it is possible to control the end of the etching based on the predicted end time.

For example, there is a certain time lag from when the supply of the etching liquid from the etching liquid supply unit 24 is stopped by the end instruction signal to when the etching liquid on the etching surface is removed by the cleaning water. Therefore, in the control method in which the time when the thickness becomes equal to or smaller than the end point thickness is set as the end time, overetching may occur. On the other hand, if the end time predicted in advance by the thickness change line is used and the supply of the etchant is stopped at a time earlier than the end time by the time lag, the overetching does not occur. Accurately control the end thickness of semiconductor wafer W

27 You can control.

The thickness measuring apparatus, the thickness measuring method, the wet etching apparatus and the wet etching method using the same according to the present invention are not limited to the above-described embodiments, and various modifications of the configuration and change of the process are possible. is there. For example, the holding substrate 21 is for maintaining the mechanical strength of the thinly etched semiconductor wafer W. Depending on the thickness of the semiconductor wafer W, the etching can be performed without using the holding substrate. It is.

Further, as for the light input means for taking in the reflected light from the semiconductor wafer W, the probe head 13 which is the light output means is shared in the above embodiment, but the light input means is separate from the light output means. May be installed. In this case, the reflected light is input to an optical fiber different from the optical fin 13a to the probe head 13 so that another optical power bra provided in addition to the optical power bra 12 is used. The reflected light and the reference light are coupled by using the light coupling means. Further, it is also possible to adopt a configuration in which only one of the optical input / output means or the optical branching and Z-coupling means is used as a single optical input / output means or optical power blur, and the other is separated.

エッチング The etching rate of the etching does not necessarily have to be constant. For example, based on the time variation of the thickness of the semiconductor wafer W obtained by the thickness measurement, if the etching is controlled so that the etching rate becomes slower as the end point thickness (etching end time) approaches, more fine control of the thickness can be achieved. Becomes possible. In this case, it is also possible to divide the time range of the raw thickness value from which the time change of the thickness is obtained, and to obtain the thickness change straight lines before and after the time when the etching rate is changed.

In addition, each calculation of the statistical process for calculating the thickness change straight line and the statistical thickness value may be variously modified in addition to the above-described embodiment. For example, at the first measurement time after the lapse of the specified time, the de-night selection calculation performed before the determination of the thickness change line is performed twice in the above embodiment. Overnight

28 The screening may be done only once or three or more times.

Industrial applicability

A thickness measuring apparatus, a thickness measuring method, a wet etching apparatus using the same, and a wet etching method according to the present invention include two lights selected from a light intensity distribution of interference light that combines reflected light and reference light. In addition to calculating the raw thickness value of the semiconductor wafer W from the difference in the optical path length of the intensity peak, statistical processing including judgment outside the allowable numerical value range and determination of the thickness change straight line, in particular, at the first measurement time after the specified time has elapsed, Performing statistical processing, including overnight selection calculations and determination of the thickness change line, makes it possible to measure the thickness of semiconductor wafers irrespective of the presence of an etchant. The present invention can be used as a thickness measuring device and a method capable of obtaining a sufficiently reduced statistical thickness value.

By using such thickness measurement, it is possible to know the actual etching rate and the etching conditions such as its time change in each wet etching process by actual measurement. Therefore, after the end of the wet etching, it is not necessary to judge the quality of the etching at the inspection stage for measuring the thickness of the obtained semiconductor wafer, but to perform the wet etching while judging the time change of the thickness during the etching. As a result, the efficiency of semiconductor device manufacturing and the yield can be improved.

29

Claims

The scope of the claims

1. A thickness measuring device for measuring the thickness of a semiconductor wafer during gate etching using an etching solution,

A measurement light source for supplying a measurement light at each of a plurality of measurement times at predetermined time intervals;

Light branching means for branching the measurement light from the measurement light source,

A light output unit that outputs one of the measurement lights branched by the light branching unit to the semiconductor wafer to be measured, and irradiates the semiconductor wafer from the etching surface side to which the etching liquid is supplied;

Light input means for inputting reflected light of the measurement light irradiated from the light output means reflected by the etching liquid or the semiconductor wafer;

The other of the measurement light branched by the light branching unit, the reference light generation means for generating a reference light having a reference light path length set by passing a reference light path having a variable optical path length,

The reflected light from the light input unit; and the reference light from the reference light generation unit.

'An optical coupling means for coupling

Light detection means for detecting the interference light from the light coupling means,

At each of the measurement times, it is set by using a light intensity distribution indicating a correlation between the reference light path length set by the reference light generation means and the light intensity of the interference light detected by the light detection means. Calculating a raw thickness value of the semiconductor wafer based on an optical path length difference of the reference optical path length between two light intensity peaks selected from a plurality of light intensity peaks having a light intensity greater than the threshold value. Thickness value calculating means,

At each of the measurement times after a specified time has elapsed from the first measurement time, with respect to the time change of the plurality of raw thickness values that are valid within the set allowable numerical value range, Statistical thickness value calculating means for determining a thickness change line by a straight line approximation calculation and calculating a statistical thickness value from the thickness change line,

30 The statistical thickness value calculating means, at the first measurement time after the specified time has elapsed from the first measurement time,

For the time change of the effective raw thickness value up to the measurement time, determination of a thickness change straight line for sorting by a linear approximation calculation, setting of a selection numerical value range from the thickness change straight line for sorting, After performing the data selection calculation including the selection of the data that invalidates the raw thickness value that is outside the selection numerical range, the linear change is performed with respect to the time change of the raw thickness value that is valid after the selection. A thickness measuring device, wherein the thickness change line is determined by an approximate calculation, and the allowable numerical range is set from the thickness change line.

2. The statistical thickness value calculating means, for each of the second and subsequent measurement times after the specified time has elapsed from the first measurement time,

Depending on whether the raw thickness value calculated at the measurement time is within or outside the allowable numerical range from the thickness change line determined at the previous measurement time, the raw thickness value To determine the validity or invalidity of

When the raw thickness value is within the allowable numerical value range and is valid, with respect to a time change of a plurality of valid raw thickness values including the raw thickness value, the linear approximation calculation is used. Determine the thickness change line, calculate the statistical thickness value from the thickness change line,

When the raw thickness value is out of the permissible numerical range and is invalid, the thickness change line determined at the previous measurement time is determined as the current thickness change line as it is, 2. The thickness measuring device according to claim 1, wherein the statistical thickness value is calculated from a thickness change straight line.

3. The statistical thickness value calculating means performs the first measurement time after the specified time has elapsed from the first measurement time,

3. The thickness measuring device according to claim 1, wherein the allowable numerical value range is set to a numerical value range narrower than the selected numerical value range.

31

4. The statistical thickness value calculating means performs the first measurement time after the specified time has elapsed from the first measurement time,

The selection numerical range is set based on a variation value of the raw thickness value to be selected from the thickness change straight line for selection and a predetermined selection constant,

The said permissible numerical value range is set based on a variation value from the said thickness change straight line of the raw thickness value validated after sorting, and a predetermined permissible constant. The thickness measuring device according to any one of claims 1 to 4.

5. The statistical thickness value calculating means invalidates the raw thickness value when the number of the light intensity peaks is less than three at each of the measurement times. The thickness measuring device according to any one of claims 1 to 4.

6. The statistical thickness value calculating means performs the measurement at each of the second and subsequent measurement times after the specified time has elapsed from the first measurement time,

The thickness measuring device according to any one of claims 1 to 5, wherein the raw thickness value outside the range of the specified time from the measurement time is invalidated.

7. A wet etching apparatus comprising the thickness measuring apparatus according to any one of claims 1 to 6,

An etching liquid supply unit configured to supply the etching liquid to the etching surface of the semiconductor wafer to be subjected to the etching;

Etching control means for controlling the supply of the etching liquid by the etching liquid supply means;

A cutting device, comprising:

8. The statistical thickness value calculating means of the thickness measuring device obtains an end time of the etching from the determined thickness change line based on a preset end point thickness, and indicates the end time. Outputting an end instruction signal, wherein the etching control means, based on the end instruction signal,

32 8. The wet etching apparatus according to claim 7, wherein the supply of the etching liquid by a supply unit is stopped.

9. A thickness measurement method for measuring the thickness of a semiconductor wafer during gate etching using an etchant,

A measuring light supply step for supplying measuring light from a measuring light source at each of a plurality of measuring times at predetermined time intervals;

An optical branching step for branching the measurement light from the measurement light source;

A light output step of outputting one of the measurement lights branched in the light branching step to the semiconductor wafer to be measured and irradiating the semiconductor wafer from the etching surface side to which the etching liquid is supplied;

A light input step of inputting reflected light reflected by the etching liquid or the semiconductor wafer to the measurement light irradiated in the light output step;

A reference light generating step of passing the other of the measurement light branched in the light branching step through a reference optical path having a variable optical path length to generate reference light having a set reference optical path length;

An optical coupling step in which the reflected light input in the optical input step and the reference light generated in the reference light generation step are combined into interference light;

A light detecting step of detecting the interference light combined in the light combining step; at each of the measurement times, the reference light path length set in the reference light generating step; and the light detected in the light detecting step. The reference optical path length between two light intensity peaks selected from a plurality of light intensity peaks having a light intensity greater than a set threshold value using a light intensity distribution indicating a correlation with the light intensity of the interference light. A raw thickness value calculating step of calculating a raw thickness value of the semiconductor wafer based on the optical path length difference of the semiconductor wafer; and setting of each of the measurement times after a specified time has elapsed from the first measurement time With respect to the time change of the plurality of raw thickness values that are valid within the allowable numerical range, a thickness change straight line is determined by linear approximation calculation, and the thickness change line is determined.

33 A statistical thickness value calculating step of calculating a statistical thickness value from the change line.In the statistical thickness value calculating step, at the first measurement time after the specified time has elapsed from the first measurement time,

For the time change of the effective raw thickness value up to the measurement time, determination of a thickness change straight line for sorting by a linear approximation calculation, setting of a selection numerical value range from the thickness change straight line for sorting, After performing data selection calculation including data selection that invalidates the raw thickness value that is out of the selected numerical value range, with respect to a time change of the raw thickness value that is valid after the selection, A thickness measurement method comprising: determining the thickness change line by the straight line approximation calculation; and setting the allowable numerical range from the thickness change line.

10. In the statistical thickness value calculation step, each of the second and subsequent measurement times after the specified time has elapsed from the first measurement time,

If the raw thickness value is out of the permissible numerical value range and is invalid, the thickness change line determined at the previous measurement time is determined as the current thickness change line as it is, 10. The thickness measurement method according to claim 9, wherein the statistical thickness value is calculated from a thickness change straight line.

11. In the statistical thickness value calculation step, at the first measurement time after the specified time has elapsed from the first measurement time,

Setting the permissible numerical range to a numerical range narrower than the selected numerical range.

34 The thickness measuring method according to claim 9 or 10, wherein the thickness is measured.

1 2. In the statistical thickness value calculation step, at the first measurement time after the specified time has elapsed from the first measurement time,

The said allowable numerical value range is set based on the variation value from the said thickness change straight line of the raw thickness value validated after sorting, and a predetermined allowable constant. 3. The thickness measuring method according to any one of the above items 1.

13. In the statistical thickness value calculating step, when the number of the light intensity peaks is less than 3 at each of the measurement times, the raw thickness value is invalidated. Item 13. The thickness measuring method according to any one of Items 9 to 12.

14. In the statistical thickness value calculation step, each of the second and subsequent measurement times after the specified time has elapsed from the first measurement time,

The thickness measuring method according to any one of claims 9 to 13, wherein the raw thickness value outside the range of the specified time from the measurement time is invalidated.

15. An etching method including the thickness measuring method according to any one of claims 9 to 14, wherein

An etching start step of supplying the etching solution to the etching surface of the semiconductor wafer to be subjected to the wet etching to start the wet etching;

A thickness measurement step of measuring the thickness of the semiconductor wafer by the thickness measurement method at each of the plurality of measurement times at the time intervals during the execution of the wet etching started in the etching start step; When,

An etching ending step of stopping the supply of the etching solution and ending the etching.

35 (2) An etching method.

16. An end time calculating step of obtaining the end time of the etching based on the preset end point thickness from the thickness change line determined in the thickness measuring step,

16. The method according to claim 15, wherein, in the etching ending step, the supply of the etching liquid is stopped based on the ending time obtained in the ending time calculating step.

17. The end time calculation step, wherein the measurement time at which the statistical thickness value calculated from the thickness change straight line is equal to or less than the end point thickness is set as the end time.ゥ Etching method.

18. The jet etching method according to claim 16, wherein in the end time calculating step, the end time is predicted using the thickness variation straight line.

36